0

I have an offline system (a robot running Ubuntu, not connected to any network) I want to monitor the resource usage over time (mainly CPU and memory). I am used to Influx and Prometheus and i was wondering if there was a tool to save the monitoring data and store until I connect to the machine and export the data, and if possible export it to an online prometheus or influx instance.

I've looked at Promequeen, but it seems outdated and creates a new prometheus database instead of appending the data to one.

Are there some tools or an industry standard for this scenario?

Improve this question
4
  • What do you mean by offline, turned off or disconnected from regular Network ? if turned off, you can probably use Wake on LAN, disconnected should use a gateway from SCADA Network to Regular and it can work as usual by forwarding the correct port. Commented Jul 11, 2023 at 13:08
  • By offline I mean running but not connected to the internet (or any LAN whatsoever). Commented Jul 11, 2023 at 13:16
  • No way for disconnected machines and that's why i'm using the so-called SCADA Networks which are isolated except certain flows through NAT & GW. Commented Jul 11, 2023 at 13:25
  • The robot is a wireless mobile robot and is transported from a working site to another every day, thus cannot have dangling cable and cannot rely on on-site equipment. Network, SCADA or not, is not an option for us. Commented Jul 11, 2023 at 13:44

2 Answers 2

Reset to default
0

I'd use Prometheus Pushgateway, it store metrics until deleted and push metrics directly to Prometheus on demand.

Improve this answer
4
  • There would be two issues if I understand correctly how it works: 1. If I scrape every two day, I would have a data point only every two day, and I'll lose the data in between. 2. I would need a push-compatible exporter. For instance, I'm used to cadvisor to monitor container CPU and memory usage, but it doesn't support pushing. Commented Jul 12, 2023 at 7:05
  • ad 1. What do you mean by "I'll lose the data in between"? Push gateway store all the data unless deleted. Ad 2. Yes, you would. Commented Jul 12, 2023 at 12:13
  • I would have the last data pushed to the gateway. For example, if I push the cpu_time for each process during two days and then scrape the data, prometheus would only show the last data point. That would be the total cpu_time for every process ran within those two days, but I would not have access to the variations between the start and the end of each process. Commented Jul 12, 2023 at 12:24
  • I see. I haven't realized that the metrics in PG doesn't accumulate as set of values. Sorry for that. Commented Jul 12, 2023 at 13:46
0

In the end I've given up using Prometheus, which without a tool like Promqueen isn't made to receive out-of-date data.

I set up Telegraf on the robot and made it output the data in Influx line protocol, so I just have to retrieve the metrics file and import it into a Influx database.

Improve this answer

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.